Fuzzy Logic-Based Fall Detection System for Elderly Using a Single Inertial Measurement Unit

Accidental Falls are common occurrences among elderly people that can lead to serious injuries. Wearable fall detection systems are widely used to help detect and monitor the elderly during their Activity of Daily Living (ADLs). This paper presents a fall detection system using fuzzy logic algorithm based on a wearable inertial measurement unit (IMU) sensor placed on the user's waist. The main system is constructed using master-slave architecture. ESP-NOW protocol is implemented for the master-slave wireless data transmission. We calculate the magnitude of total acceleration based on IMU's three axes, which are used to trigger the fuzzy logic algorithm. The fuzzy logic is used to detect falls and determine fall position based on the pitch and roll angular position of the waist. Experiment is conducted on two subjects, which we get a total accuracy of 94.2%, precision of 97.14%, and recall of 97.13%. Based on our analysis, the error occurred because the user found it hard to demonstrate a natural fall position even when using protective gear and a mattress. Still, this device is able to achieve high accuracy with a low computational load, so the algorithm can be implemented on a low-cost chip or MCU.